Elucidating Parasitic Currents in Proton-Exchange-Membrane Electrolytic Cells via Physics-Based and Data-Driven Modeling

Proton-exchange membrane (PEM) water electrolysis is pivotal for green hydrogen production, necessitating accurate predictive models to manage their non-linearities and expedite commercial deployment. Understanding degradation mechanisms through macro-scale modeling and uncertainty quantification (UQ) is crucial for advancing this technology via efficiency enhancement and lifetime extension. This study primarily utilizes a one-dimensional physics-based model to elucidate the presence of electron transport within the PEM, another degradation phenomenon, besides gas crossover. This work also applies a machine learning (ML) algorithm, such as eXtreme Gradient Boosting (XGBoost), to model PEM electrolytic cell (PEMEC) operation based on a dataset generated from the previously mentioned physics-based model. The ML model excels in predicting the polarization behavior. Based on this surrogate model, UQ and sensitivity analysis are finally employed to enlighten the dependence of PEMECs performance and Faradaic efficiency on the effective electronic conductivity of PEM, especially when electronic pathways exist within the membrane and operating at low current densities.